MSL Technical Guide 19

Measuring Atmospheric
Pressure with a Barometer

Barometric Pressure Units

Summary The pascal (Pa) is the metric (SI) unit of pressure
This technical guide explains how to use a digital ba- and is defined as a force of 1 newton acting over an area
2
rometer to measure atmospheric pressure in a labora- of 1 square metre so 1 Pa = 1 N/m [1]. One pascal is a
tory or industrial setting. It starts with a brief discussion small pressure so measurements are usually reported in
of the atmosphere and then talks about barometers, their kPa (1000 Pa) or MPa (1000 kPa).
stability and calibration requirements, along with how to Barometric pressure measurements are often re-
work out the uncertainty of an air pressure measure- ported in hectopascal (hPa). Hecto means times 100 so
ment. 1 hPa = 100 Pa. The prefix hecto is used in barometry
as the numerical value of the pressure in hPa is the
same as that measured in millibar (mbar),
Introduction
1 mbar = 1 hPa = 100 Pa. Many historical meteorological
A barometer measures the air pressure of the Earth’s records of atmospheric pressure were recorded in mbar.
atmosphere. The atmospheric pressure is generated by There are many other barometric pressure units,
the Earth’s gravity acting on the mass of air in the at- some of which are shown in the table below, along with
mosphere. The forces involved are surprisingly large, a their symbol and relationship to the pascal.
surface pressure of 100 kPa means the force at the base
2
of a column of the atmosphere, cross-section area 1 m ,
is equivalent to a mass of 10,000 kg. The atmospheric
pressure depends on the local environmental conditions Name Symbol Relation to SI
such as air temperature, altitude, and weather pattern.
pascal (SI unit) Pa N m-2 = kg m-1 s-2
A typical atmospheric pressure measurement, made
in New Zealand at sea level, will be in the range 970 hPa hectopascal hPa 1 hPa = 100 Pa
to 1040 hPa. The graph below shows the air pressure
recorded over a month. bar bar 1 bar = 100 000 Pa

millibar mbar 1 mbar = 100 Pa

torr torr 1 torr ~ 133.322 Pa

1035
1030 millimetre of mercury
mmHg 1 mmHg ~ 133.322 Pa
(conventional)
1025
pounds per square
Pressure / hPa

1020 psi 1 psi ~ 6894.757 Pa

1015
1010
1005
1000
995
990
0 10 20
Time / day
A barometric pressure measurement is an example
of an absolute pressure measurement. This means a
barometer would read zero when measuring a vacuum
(zero pressure absolute). Most pressure instruments,
such as a car tyre gauge, measure gauge pressures, so
read zero when open to the atmosphere.
Barometric pressure measurements are used in a
wide variety of situations such as weather forecasting,
air density calculation, measuring changes in altitude
and in a range of thermodynamic calculations such as
gas turbine efficiency.
inch
Pressure units for liquid manometers, like millimetre
of mercury or inch of water, were commonly used. This
type of unit should not now be used, as it is obsolete and
can cause confusion. The problem is that “manometric”
pressure units can have more than one definition.
30
Types of Barometer
Barometers come in many different types, ranging
from mercury filled glass instruments, mechanical aner-
oid instruments, to resonant sensors made using silicon
fabrication technology. We will briefly discuss digital ba-
rometers and end this section with a warning about why
mercury-in-glass barometers should not be used.
Digital Barometers
All digital barometers have a pressure transducer
that converts the force generated by the pressure into an
electrical signal. The most common type of transducer is
a thin metal diaphragm with the atmospheric pressure on
one side and a vacuum on the other. The changing at-

Measurement Standards Laboratory of New Zealand • fax: +64 (0)4 931 3003 • e-mail: msl@irl.cri.nz • website: http://msl.irl.cri.nz 1
mospheric pressure deflects the diaphragm. This deflec-
tion is measured either directly or as a change in dia-
phragm tension.
Currently the best performing barometers are based
on silicon or quartz resonant sensors whose frequency
depends on the applied pressure.
Mercury Barometers
MSL does not offer a calibration service for mercury
barometers. Mercury is hazardous to human health and
is a difficult fluid to handle and keep clean. Mercury-in-
glass barometers are difficult to use and the readings
always require significant correction. We strongly rec-
ommend replacing mercury-in-glass instruments with
digital barometers. Mercury barometers must be dis-
posed of carefully through an approved waste disposal
company.
Barometer Selection
Factors to consider, when selecting a barometer, in-
clude the required measurement uncertainty, stability
(drift rate), ease and frequency of calibration and fea-
tures such as a computer interface.
Measurement Uncertainty
Most digital barometers will easily achieve meas-
urement uncertainties of < 1 hPa. This will be adequate
for most applications; for example weather monitoring
and forecasting need to resolve pressure changes of
less than 1 hPa while air density calculations require un-
certainties of 10 hPa or less [2]. Here at MSL we can
provide barometer calibrations with expanded uncertain-
ties as low as 0.02 hPa, providing the barometer is ca-
pable of performing at this level.
Barometer Stability
Barometer stability is the drift in measured pressure
with time. Stabilities of < 0.1 hPa per year are achiev-
able although this will depend on the barometer. Some
barometers that we have calibrated over several years,
show drift rates as low as 0.02 hPa/yr. Select a barome-
ter with an annual drift rate that is less than your re-
quired measurement uncertainty.
Cost of Calibration
Barometer recalibration intervals range between six
months and two years depending on the instrument, the
required uncertainty and the conditions of use. Barome-
ters with a lower drift rate generally have a longer recali-
bration interval so have a smaller operating cost.
Choose a barometer with a pressure port, i.e. a tube
allowing a pressure tight connection. Instruments without
a pressure port need to be placed in a chamber, which
complicates the calibration and increases the cost.
Computer Interface
Choose a barometer with a computer or instrument
interface, such as USB, LAN, RS232, etc. An interface
allows the atmospheric pressure to be read automati-
cally and logged by computer. This is particularly useful
when measuring changing air pressures.
Measurement Standards Laboratory of New Zealand • fax: +64 (0)4 931 3003 • e-mail: msl@irl.cri.nz • website: http://msl.irl.cri.nz
Atmospheric Pressure Measurement
An atmospheric pressure measurement can be as
simple as just recording the reading on the instrument
display, as long as the following conditions are met. The
barometer used for the reading should be correctly in-
stalled and have a calibration certificate from an accred-
ited calibration laboratory. Performance checks are
needed to ensure the barometer is working correctly be-
tween calibrations.
Barometer Installation
Follow the manufacturer’s recommendations but also
take into account how the instrument was calibrated.
Factors to consider include:
• Instrument orientation. Some barometers are sensi-
tive to tilt and are normally oriented upright and
level. Ensure that the barometer is calibrated in the
orientation in which it is used.
• Measurement conditions. Drafts and flow of air
across the pressure port can cause small meas-
urement errors. Temperature sensitivity may be a
factor for barometers mounted out-of –doors.
• Head correction. Remember that the barometric
pressure changes with altitude, it decreases by
~1.2 hPa for a 10 m increase in attitude (near sea
level). Liquid trapped in the pressure port or tubing
will also cause errors, a 10 mm height of water will
produce an error of ~1 hPa.
You may need to apply the calibration corrections to
the barometer reading although often these corrections
are small enough that they can be ignored. The correc-
tions in a MSL calibration report are simply added to the
displayed reading. For example if the barometer is dis-
playing a pressure reading of 980.25 hPa and the cali-
bration report correction at 980 hPa is –0.1 hPa then the
corrected pressure reading is
p = 980.25 + ( −0.1) = 980.15 hPa .
Measurement Uncertainty
Calculating the uncertainty of an air pressure meas-
urement is straightforward. There are only two or three
uncertainty terms that need to be considered, most of
which are known before any measurement is made. The
first two terms arise from the barometer performance;
they are the calibration uncertainty and hysteresis. The
third term is the measurement repeatability that arises
from the barometer performance and from the stability of
the pressure being measured. See MSL Technical Guide
13 for more discussion of pressure gauge uncertainty
[1].
Calibration Uncertainty
The calibration uncertainty is taken directly from the
barometer calibration report. A typical MSL calibration
report uncertainty statement gives the expanded uncer-
tainty of the corrections Ucal at a 95 % confidence level,
refer GUM [3]. We can convert from expanded to stan-
dard uncertainty by dividing by the coverage factor k. For
example, if Ucal = 0.06 hPa and k = 2.2 then
Ucal 0.06
ucal = = = 0.027 hPa .
k 2.2
2
Here the uppercase symbol U refers to expanded uncer-
Time / s Reading / hPa Spreadsheet Function
tainty while u refers to the standard uncertainty.
1 1015.25
Hysteresis 2 1015.10
3 1015.25
The hysteresis uncertainty component can be esti-
4 1015.25
mated from the difference between the calibration report 5 1015.30
rising and falling pressure corrections, at the same 6 1015.30
nominal pressure. The hysteresis is normally negligibly 7 1015.05
small for digital barometers. 8 1015.20
If it is necessary to calculate the hysteresis uncer- 9 1015.30
tainty first find the largest difference, dmax, between the 10 1015.05
rising and falling pressure corrections (at the same Average: 1015.21 =AVERAGE(C2:C11)
nominal pressure). This is normally at a midpoint calibra-
tion pressure. Then estimate the standard uncertainty Standard deviation: 0.10 =STDEV(C2:C11)
component using uhys = dmax (2 3) (here we have
treated the hysteresis uncertainty as a rectangular distri-
bution with a half width of dmax 2 [3]).
For example, if dmax = 0.15 hPa then The standard uncertainty due to the repeatability urep
uhys = 0.15 (2 3) = 0.04 hPa. is just the standard deviation; for the example above.
urep = 0.10 hPa.
Repeatability
Resolution
The repeatability uncertainty component is an esti-
mate of how well successive pressure measurements If urep = 0 hPa, meaning all ten measurements where
agree. If the measured pressure is stable then this term the same, then you need to replace urep with a term for
is just the repeatability of the barometer. If the measured the gauge resolution uncertainty ures. For example if the
pressure is unstable then there will be an uncertainty resolution of the barometer is 0.05 hPa then
component associated with the pressure fluctuations.
0.05
ures = = 0.014 hPa.
2 3
1016.19
1016.18
Finally, we can obtain the uncertainty in the meas-
1016.17
ured pressure by combining all the terms together using
Pressure / hPa

1016.16 the equation

1016.15
1016.14
1016.13
uc = (u 2
cal
2
+ uhys 2
+ urep )
1016.12
1016.11 (if urep = 0 then substitute ures).This equation means you
1016.10 square all the terms, add them together, and take the
1016.09 square root of the sum.
0 200 400 600 800 1000 The expanded uncertainty for the pressure reading is
Time / second then calculated using Uc = kuc ≈ 2.2uc . Here we have
made the approximation of setting k = 2.2. The actual
coverage factor can be calculated but is unlikely to differ
In practice, the short-term fluctuations in the atmos- from 2.2 by more than 10 percent.
pheric pressure are small. The graph above is the The result for our worked example is
measured atmospheric pressure in a laboratory recorded
over 15 minutes. It shows a variation of about 0.1 hPa uc = (0.0272 + 0.042 + 0.12 )0.5 = 0.11 ≈ 0.10 hPa .
due to the weather pattern and some short-term fluctua-
tions of ~ 0.01 hPa due to drafts caused by the opening
and closing of doors etc. Compare this with the barometer resolution of 0.05 hPa.
The repeatability uncertainty component can be as- The expanded uncertainty is given by
sessed by calculating the standard deviation of succes-
sive pressure readings. This calculation should be done Uc = 2.2 × 0.11 = 0.24 ≈ 0.25 hPa.
to test each new measurement setup. The resulting un-
certainty component will normally be small and close to For a single barometer reading we would report a read-
the resolution uncertainty. ing of 1015.20 hPa with an expanded uncertainty of
On the next column is a spreadsheet example of a 0.25 hPa.
repeatability uncertainty calculation. Ten pressure read-
ings were recorded, at 1 second intervals, with a ba-
rometer that has a resolution of 0.05 hPa. The pressure
is fluctuating in the range 1015.0 hPa to 1015.3 hPa.
The bottom two rows of the table show the calculated
average pressure and standard deviation. The third col-
umn shows the worksheet functions used (for data in
rows 2 to 11 of column C of the spreadsheet).

Measurement Standards Laboratory of New Zealand • fax: +64 (0)4 931 3003 • e-mail: msl@irl.cri.nz • website: http://msl.irl.cri.nz 3
References Further Information
[1] MSL Technical Guide13: “Pressure Gauge Calibra- If you want to know more, about barometers then
tion”, http://msl.irl.cri.nz. contact MSL via the MSL website: http://msl.irl.cri.nz.
MSL offers a range of one-day measurement workshops
[2] MSL Technical Guide 7: “Calibrating Standard covering practical aspects of electrical, density, mass,
Weights”, http://msl.irl.cri.nz. pressure and temperature measurement and measure-
[3] Guide to the Expression of Uncertainty in Measure- ment uncertainty, see the website for more details.
ment, ISO, 1995.

Prepared by Mark Fitzgerald and Darrin Jack, June

2008.

MSL is New Zealand’s national metrology institute, operating within Industrial Research Limited under the authority of the
New Zealand Measurement Standards Act 1992.

Measurement Standards Laboratory of New Zealand • fax: +64 (0)4 931 3003 • e-mail: msl@irl.cri.nz • website: http://msl.irl.cri.nz 4